12
Commissioning: Commissioning: Preliminary thoughts from Preliminary thoughts from the Offline side the Offline side ATLAS Computing Workshop ATLAS Computing Workshop December 2004 December 2004 Rob McPherson Rob McPherson Hans von der Schmitt Hans von der Schmitt

Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

Embed Size (px)

Citation preview

Page 1: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

Commissioning: Commissioning:

Preliminary thoughts from the Preliminary thoughts from the Offline side Offline side

ATLAS Computing WorkshopATLAS Computing Workshop

December 2004December 2004

Rob McPhersonRob McPherson

Hans von der SchmittHans von der Schmitt

Page 2: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

2

ATLAS CommissioningATLAS Commissioning Commissioning = Commissioning = “Just Installed” “Just Installed” “Operational” “Operational” Referring primarily to activities at Point 1 (+ integration)Referring primarily to activities at Point 1 (+ integration) Broken into 4 phasesBroken into 4 phases

Phase 1: subsystem standalone commissioningPhase 1: subsystem standalone commissioning DCS: LV, HV, cooling, gas, safety systems, record in DB, retrieve from DBDCS: LV, HV, cooling, gas, safety systems, record in DB, retrieve from DB DAQ: pedestal runs, electronic calibration, write data, analyzeDAQ: pedestal runs, electronic calibration, write data, analyze

Phase 2: integrate systems into full detectorPhase 2: integrate systems into full detector Phase 3: cosmic raysPhase 3: cosmic rays

Take data, record, analyze/understand them, distribute to remote sitesTake data, record, analyze/understand them, distribute to remote sites Phase 4: single beam, 1Phase 4: single beam, 1stst collisions collisions

Same, with increasingly higher ratesSame, with increasingly higher rates

Phases will overlapPhases will overlap Some systems may take cosmics while others are still installingSome systems may take cosmics while others are still installing

Starts very soonStarts very soon Barrel calos start “phase 1” electronics commissioning Barrel calos start “phase 1” electronics commissioning Mar 2005 Mar 2005

Page 3: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

3

ATLAS Commissioning StructureATLAS Commissioning Structure

CryogenicsCryogenicsG. Passardi

Det. cooling, GasDet. cooling, GasJ. Godlewski

Cooling,VentilationCooling,VentilationB. Pirollet

SafetySafetyG. Benincasa

MagnetsMagnetsH. Ten Take

DatabasesDatabasesR. Hawkings, T. Wenaus

OfflineOfflineR. McPhersonH. von der Schmitt

PixelsPixelsL. Rossi

LARLARL. Hervas

TilesTilesB. Stanek

Mu-BaMu-BaL. Pontecorvo

Mu-ECMu-ECS. Palestini

Lumi.,Beam-pipeShieldings,Etc..SCTSCT

S. McMahon

TRTTRTH.Danielsson

P. Lichard

IDIDP. Wells

DAQDAQG. Mornacchi

CentralCentralDCSDCS

H. BurckhartLVL1LVL1T. Wengler

HLTHLTF. Wikkens

TDAQTDAQG.Mornacchi

TheThe names are are the current the current commissioning commissioning contact peoplecontact people

OVERALL ATLASOVERALL ATLASG.Mornacchi, P.Perrodo

Page 4: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

4

Offline commissioningOffline commissioning Work program will start with detector debugging/monitoring in Work program will start with detector debugging/monitoring in

early stages, move to cosmics, beam halo, beam gas, and then early stages, move to cosmics, beam halo, beam gas, and then 11stst collisions collisions

Many issues for detector online+offline software, databases, Many issues for detector online+offline software, databases, simulation, data distribution, remote reconstruction, ...simulation, data distribution, remote reconstruction, ...

Will have meetings as neededWill have meetings as needed Request contact people from detectors and related groups with Request contact people from detectors and related groups with

some responsesome response ID : Maria Costa LAr : ? Tiles : Sasha Solodkov Muons : ? DB : Richard Hawkings and Torre Wenaus Simulation : ? Physics : ?

Page 5: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

5

RCCRCCVMEVME

Detector commissioning: offline viewDetector commissioning: offline view

DETECTORDETECTOR

??

ConfigurationConfigurationDatabase(s)Database(s)

ConditionsConditionsDatabase(s)Database(s)

DCS / ControlsDCS / Controls HV, LVHV, LV Temp sensorsTemp sensors AlignmentAlignment CryogenicsCryogenics ......

TDAQ/systemTDAQ/systemWorkstationWorkstation

(GNAM in CTB)(GNAM in CTB)

OnlineOnlineSystemSystem

PresenterPresenter

OnlineOnlineHisto SvcHisto Svc??

Front-EndFront-End

RODROD

ROSROS

Bytestream Bytestream FilesFiles

SFISFI

EFEF

SFOSFO

LVL2LVL2

LVL1LVL1

ATHENAATHENA

Page 6: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

6

non-Event data accessnon-Event data access DCS and other controls dataDCS and other controls data

If needed offline, natural access via conditions DB interfaceIf needed offline, natural access via conditions DB interface Can we assume evolution of current PVSS manager sufficient? Can we assume evolution of current PVSS manager sufficient?

Probably yes. Assume it will move to Oracle at some point.Probably yes. Assume it will move to Oracle at some point.But must watch custom DB use in case additional central support tools But must watch custom DB use in case additional central support tools

required.required.And must also watch data volume into relational DB ...And must also watch data volume into relational DB ...

Note that most “DCS” monitoring tasks is done in PVSS et al. Note that most “DCS” monitoring tasks is done in PVSS et al. (Not called “offline” here)(Not called “offline” here)

Configuration informationConfiguration information Again, assume necessary information written either to event stream or Again, assume necessary information written either to event stream or

to conditions DBto conditions DB

Calibration / alignment constantsCalibration / alignment constants Obviously need conditions DB for theseObviously need conditions DB for these

Will we have common system in time for commissioning ? Will we have common system in time for commissioning ? (POOL et al?)(POOL et al?)

Can fallback to CTB systems ... not very nice ....Can fallback to CTB systems ... not very nice ....

Page 7: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

7

Event Data Access (1)Event Data Access (1)1)1) Running ATHENA on ByteStream / EventStorageFilesRunning ATHENA on ByteStream / EventStorageFiles

Easiest way for offline code to access dataEasiest way for offline code to access data Would we want to maintain a “commissioning branch” like the Would we want to maintain a “commissioning branch” like the

CTB?CTB? Would we want this branch built on non-afs like the CTB?Would we want this branch built on non-afs like the CTB?

Can re-use a lot of monitoring tools developed for CTBCan re-use a lot of monitoring tools developed for CTB Ideally ROD Ideally ROD ROB ROB ROS chain working, but can also RCC ROS chain working, but can also RCC BS BS

Limited number of channels: ROB/ROS Limited number of channels: ROB/ROS PC with Filar card PC with Filar card (need for subdetectors without full VME readout)(need for subdetectors without full VME readout)

2)2) In Event FilterIn Event Filter Requires more of the DAQ system runningRequires more of the DAQ system running Experience from CTB : not always possible to keep code up-to-dateExperience from CTB : not always possible to keep code up-to-date

uncouple detector monitoring from online software releases etc. as uncouple detector monitoring from online software releases etc. as much as possible?much as possible?

Need to review handling of “incidents” (asynchronous interrupts) Need to review handling of “incidents” (asynchronous interrupts) passed into the ATHENA job passed into the ATHENA job

Histogram reset under certain conditions ...Histogram reset under certain conditions ...

Page 8: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

8

Event Data Access (2)Event Data Access (2)3)3) ““Online” workstation (a la GNAM in the CTB)Online” workstation (a la GNAM in the CTB)

Can take (ethernet) data stream via ROS (or RCC) ?Can take (ethernet) data stream via ROS (or RCC) ? Need to review this ethernet data stream and how to read it Need to review this ethernet data stream and how to read it

from an ATHENA jobfrom an ATHENA job Also need to review running ATHENA on lower level (ROD?) Also need to review running ATHENA on lower level (ROD?)

fragment (is all needed information available?)fragment (is all needed information available?) If we run ATHENA here, require:If we run ATHENA here, require:

Possibility of “light-weight” ATHENA with only converters Possibility of “light-weight” ATHENA with only converters and histogrammingand histogramming

If we don’t run ATHENA here:If we don’t run ATHENA here: Require duplication of converters and parallel maintenanceRequire duplication of converters and parallel maintenance Limited monitoring possible at this level Limited monitoring possible at this level unless we also want unless we also want

to duplicate cabling/mapping services, database interaction, to duplicate cabling/mapping services, database interaction, etc.etc.

But we will want to match histogram root tree to ATHENA in But we will want to match histogram root tree to ATHENA in any case to use same plots / macros / etc. any case to use same plots / macros / etc.

Page 9: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

9

ATHENA “online”ATHENA “online” Direct access to TDAQ Information Service (IS) essentialDirect access to TDAQ Information Service (IS) essential

Had limited use in CTB04 monitoring (eg beam energy for histograms)Had limited use in CTB04 monitoring (eg beam energy for histograms) Found this a weak point that could use reviewFound this a weak point that could use review

Need a structured monitoring/histogramming Need a structured monitoring/histogramming environment that matches online useenvironment that matches online use

Dynamic booking / rebooking of histogramsDynamic booking / rebooking of histograms Zero histograms based on some external inputZero histograms based on some external input

Eg, shift crew presses a “reset” button ... or change in some Eg, shift crew presses a “reset” button ... or change in some condition picked up via the Information Servicecondition picked up via the Information Service

Can work features into AthenaMonitoring package, once we understand Can work features into AthenaMonitoring package, once we understand what features are wantedwhat features are wanted

Need a “state model” for online system, mapped/implemented in ATHENANeed a “state model” for online system, mapped/implemented in ATHENA

Need a “smaller” buildNeed a “smaller” build Strong feeling on subdetector/TDAQ side that ATHENA is too hard to Strong feeling on subdetector/TDAQ side that ATHENA is too hard to

use for the “GNAM” environmentuse for the “GNAM” environment Hard to use, crashes in obscure places (say, ByteStreamSvc Hard to use, crashes in obscure places (say, ByteStreamSvc

somewhere due to corrupt events? How to debug this? It will somewhere due to corrupt events? How to debug this? It will happen a lot during commissioning!)happen a lot during commissioning!)

Page 10: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

10

Summary thoughts on toolsSummary thoughts on tools DatabasesDatabases

DCS databases and offline access seem OK for early commissioningDCS databases and offline access seem OK for early commissioning Calibration/Alignment databases need rationalizationCalibration/Alignment databases need rationalization

It would be very nice to have on recommended/supported solution before It would be very nice to have on recommended/supported solution before these are seriously required and used.these are seriously required and used.

Some CTB04 solutions (Nova, writing significant data into CDB) won’t Some CTB04 solutions (Nova, writing significant data into CDB) won’t scale scale

Want to archive histograms etc. from commissioning phase in DB??Want to archive histograms etc. from commissioning phase in DB??

If we use ATHENA-based event stream monitoringIf we use ATHENA-based event stream monitoring Many of the CTB tools can be migrated to commissioningMany of the CTB tools can be migrated to commissioning

Monitoring Algorithms/AlgTools, root macros, etc.Monitoring Algorithms/AlgTools, root macros, etc. Need to think about detailed plots etc. for full ATLASNeed to think about detailed plots etc. for full ATLAS

Have been ridiculous monitoring histogram extrapolations from CTB Have been ridiculous monitoring histogram extrapolations from CTB ATLAS ... must review theseATLAS ... must review these

There is a “histogram checker” (Monitoring/MonHighLevel from Manuel Diaz) There is a “histogram checker” (Monitoring/MonHighLevel from Manuel Diaz) framework in place, but needs clientsframework in place, but needs clients

If we also use non-ATHENA-based event stream monitoringIf we also use non-ATHENA-based event stream monitoring Surely would still want a common frameworkSurely would still want a common framework Can still recycle many root macros etc. from CTBCan still recycle many root macros etc. from CTB

Page 11: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

11

Phase 3 and beyond ...Phase 3 and beyond ... Have fully simulated cosmics, beam halo and beam Have fully simulated cosmics, beam halo and beam

gas samples available for detector studiesgas samples available for detector studies Some use so farSome use so far

Tiles : commissioning trigger rate studiesTiles : commissioning trigger rate studies Muons : tracking package for non-pointing, out-of-time eventsMuons : tracking package for non-pointing, out-of-time events LAr and ID : some rate and reconstruction studiesLAr and ID : some rate and reconstruction studies

Do we want dedicated samples with special detector Do we want dedicated samples with special detector configurations? Or more statistics of the samples we have?configurations? Or more statistics of the samples we have?

So far: only G3 simulation of overburden/cavern. Want G4?So far: only G3 simulation of overburden/cavern. Want G4? Need to review the readiness of the subdetector Need to review the readiness of the subdetector

reconstruction software for these non-standard eventsreconstruction software for these non-standard events Once we’re taking data with full TDAQ chain in placeOnce we’re taking data with full TDAQ chain in place

Data distribution to “Tier0” and remote computing centres Data distribution to “Tier0” and remote computing centres planned for cosmics and single-beam dataplanned for cosmics and single-beam data

Considering this is not currently highest priority, but must Considering this is not currently highest priority, but must keep in mind that this will run in parallel with DC3keep in mind that this will run in parallel with DC3

Page 12: Commissioning: Preliminary thoughts from the Offline side ATLAS Computing Workshop December 2004 Rob McPherson Hans von der Schmitt

2004/12/06 Rob McPherson / Hans von der Schmitt

12

SummarySummary ATLAS commissioning at point 1 starts in a few monthsATLAS commissioning at point 1 starts in a few months Initially, may need fallback DB solutions, but need to work to avoid Initially, may need fallback DB solutions, but need to work to avoid

these if possiblethese if possible Must watch data rate and volume written into relational DBMust watch data rate and volume written into relational DB

Will use ATHENA for detailed data analysisWill use ATHENA for detailed data analysis Must think about “AthenaMonitoring” environment for non-developersMust think about “AthenaMonitoring” environment for non-developers

Smaller, faster, simpler, robust ...Smaller, faster, simpler, robust ... Maybe need to define incident path to react inside ATHENA to Maybe need to define incident path to react inside ATHENA to

changing external conditions matching to TDAQ stateschanging external conditions matching to TDAQ states Also must verify that all subdetectors implement BS fragment Also must verify that all subdetectors implement BS fragment

versioning. Will evolve significantly during detector commissioningversioning. Will evolve significantly during detector commissioning Must consider if ATHENA also OK for “in the pit while plugging in a Must consider if ATHENA also OK for “in the pit while plugging in a

board” monitoring and then subsequent standard online board” monitoring and then subsequent standard online monitoringmonitoring

TDAQ event stream TDAQ event stream ATHENA ? ATHENA ? Regardless, it would still be good to maintain code in only one place Regardless, it would still be good to maintain code in only one place

Also need to review detector reco for cosmics etc.Also need to review detector reco for cosmics etc. And eventually also the best timescale to distribution of And eventually also the best timescale to distribution of

commissioning events to external computing centres commissioning events to external computing centres